[SciPy-dev] Inclusion of cython code in scipy

[Nathan Bell]

On Wed, Apr 23, 2008 at 12:01 PM, Matthew Brett <matthew.brett at gmail.com> wrote:
>
>  I really don't think there's a reason to discourage it at this point, and if it works,
> as several of us have guessed, then we have gained a great deal.

Conversely, there's no reason to blindly encourage the use of an
unproven technology.  I really *would* like Cython to succeed, but I
really *dislike* the approach of "we need to reimplement X in Cython,
because Cython is amazing".  I'm not arguing against the use of
Cython, I'm only moderating what comes across as naive evangelism for
a shiny new technology.

For example, I have heard numerous statements about how incredibly
slow SWIG wrappers are ("because the top layer is pure Python").  This
may be true, on the other hand it may not be true.  It may be true
only for some applications and not for others.  Ultimately, the burden
of proof is on the Cython proponents to provide supporting evidence
for their claims.  Once we have this evidence we can make an informed
decision about how to proceed.

-- 
Nathan Bell wnbell at gmail.com
http://graphics.cs.uiuc.edu/~wnbell/